Many companies tap systems integrators or consultants for in-depth preassessments prior to a voice-over-IP deployment, but administrators should take care to plan and budget for in-house monitoring and assessment tools to ensure continued success long after the initial deployment.
At the Interop conference earlier this month in Las Vegas, several panels of IT professionals provided in-depth insight into their companies VOIP deployments, highlighting best practices as well as missed opportunities. These Interop speakers universally agreed on the critical importance of conducting network performance preassessments before deploying VOIP, although they didnt outline steps or tools to do so.
Performance assessments suss out problems throughout the network. These problems might not noticeably affect data communications, but they can drastically degrade real-time applications such as VOIP. Poor cabling, out-of-date firmware, drivers and bandwidth congestion are among the many conditions that can torpedo voice quality over the network.
Large organizations often tap third-party integrators to perform upfront assessments, but administrators must plan to bring tools in-house for long-term troubleshooting and must foster working relationships with integrators during the preassessment to help gain in-house expertise for long-term remediation.
A dizzying variety of new voice-specific monitoring and assessment tools is coming to market. Many, such as ClearSight Networks Inc.s ClearSight Analyzer, focus on evaluating the performance characteristics of actual voice traffic. Others, such as NetIQ Corp.s Vivinet Assessor and Apparent Networks Inc.s AppareNet Voice, generate simulated traffic to provide end-to-end network assessment, gleaning performance characteristics and highlighting trouble spots before VOIP equipment ever touches the network.
All these products evaluate the delay, jitter and utilization characteristics of the network and translate the findings into assessment scales. The MOS (Mean Opinion Score) and the R-Value score are the most frequently utilized grading scales, attributing a numerical value to voice quality.
NetIQs Vivinet Assessor relies on performance endpoints distributed throughout the network to simulate call loads on the network and report performance characteristics over time. In contrast, AppareNet Voice uses ICMP (Internet Control Message Protocol) traffic from a central monitor to query IP addresses throughout the network, injecting only a small amount of traffic to remote hosts to provide a snapshot of network performance and to identify trouble spots along the route.
Each of these approaches has its pros and cons: Administrators must weigh the deployment complexities and performance degradation of live networks at potential bottlenecks against the shorter snapshot views of network performance. However, any of these products will provide insight into the things that matter most for voice quality and will help ensure a smooth transition to a converged network.
Good call, bad call
The MOS and the R-Value score are the most common scoring methods for voice quality. The compression characteristics of voice codecs limit possible scores for VOIP networks, so the highest grades are never actually achievable.
- MOS Subjective voice assessment score traditionally based on user-perception reports, with a score of 1 meaning unusable and 5 meaning excellent.
- R-Value Part of the ITU (International Telecommunication Union) G.107 specification to measure call quality. On a 1-to-100 scale, R-Value is an objective measurement technique that accounts for loudness and various signal impairments— and even includes a fudge factor to weigh for user expectations.
Technical Analyst Andrew Garcia can be reached at firstname.lastname@example.org.
Check out eWEEK.coms for the latest news, views and analysis on voice over IP and telephony.